Sunday, March 8, 2015

Yes David: Unemployment is Sometimes Involuntary

My pal David Andolfatto doesn't like it when I say that some unemployment is involuntary. Here is my response:

I am happy with the way you characterize my beliefs in the first paragraph of your blog. Unemployment is clearly not Pareto optimal.  Everything you say after that is at best misleading and at worst dismissive of everything we (at least some of us) learned from Keynes. 

The idea of involuntary unemployment was introduced by Keynes in the General Theory. But you already knew that. It is defined as a situation where (in modern language) the ratio of the marginal disutility of work to the marginal utility of consumption is not equal to the real wage. That seems a pretty accurate description of the equilibrium outcome of labor search models.

Bob Lucas cast a spell over the profession in a series of papers in the 1970s. You are accurately summarizing Bob's view. That view was tied to a three decade long campaign by economists predominately located in Chicago, Minnesota and Rochester (at the time) to discredit Keynesian economics. Tom Sargent reputedly advised his students not to read the General Theory. That was a tragic mistake and we are still suffering from the consequences.

You are right to assert that the important distinction is between equilibria that are Pareto optimal and those that are not. You are wrong to assert that the term 'involuntary unemployment' has no useful meaning. 

I accept your categorization of the allocation of time between three competing ends. Every family, and every member of that family, chooses every day whether they will choose to participate in the labor force. As long as they are in the labor force, they may be employed or unemployed. Those who are unemployed do not choose that state. They must wait for a job offer to appear. In some states, that job offer may take a couple of days to arrive. In others, it may take a couple of years. The activity of waiting for a job, even when it involves active search, can meaningfully be called involuntary unemployment.

The dismissal of 'involuntary unemployment' from the lexicon of the modern economist was introduced as part of a deliberate attack on Keynesian economics. It is time to roll back that attack. As I have shown here, 'involuntary' unemployment is a useful way of distinguishing unemployment that is part of a social optimum, from unemployment that is not.


  1. Well, first off, you did not say in your original post that "some" unemployment is involuntary. That qualifier would have been less offensive. :)

    Here is my reply to your post:

    Roger, I did not write this post to justify past challenges to Keynesian thinking. I think you're getting caught up in past wars.

    Look at this way. I understand your point of view (indeed, the mathematical models you work with make that view very plain). I have a lot of empathy for your view. My understanding and empathy are in no way enhanced by your use of the adjective "involuntary." If it is an involuntary state, how do explain the spike in exit rates at UI exhaustion? How?

    1. "I think you're getting caught up in past wars."

      Some economists gave up on the fight over the use of the term 'involuntary unemployment'. Partly this was due to exhaustion and partly an earlier generation of economists was overwhelmed by a generation of young turks (like you and me) armed with shiny new contraction mappings. The battle was lost. But not the war.

      You ask about a spike in exit rates. Do you mean that people leave the labor force when their UI benefits are exhausted? If so: that makes my point for me. The participation decision is voluntary. Or do you mean something else?

    2. Exit from unemployment (looking for work) does not equal re-employment or lack of interest in work. Some things are too obvious for economists.

    3. The spike in UI exit rates is NOT just to out of the labor force, it is also into employment. And it is *especially* into employment if the UI system is set up such that some work experience is necessary for eligibility.

      So how this makes your case is really quite beyond me. If the unemployed exit into NILF at UI expiration, it demonstrates that they had a *choice* to do so earlier. Likewise, if they exit to employment.

      I repeat, just because their choices are voluntary (and hence, respond to incentives) is in no way to be construed with forming a value judgement about their behavior or the welfare properties of an equilibrium. It's just bad language that confuses more than it clarifies.

      Did you ever, by the way, define "voluntary" unemployment and attempt to measure it in the data?

      And while your at it, please de-compose employment into its "voluntary" and "involuntary" components. I do hear of many working poor who wish they had the financing to go to school, for example.

  2. I decided to make this a 3-way Western war:

    1. Could we agree instead, to be the three musketeers? 😄 All for one and one for all?

  3. Why did unemployment skyrocket around 2008-2009 when the housing bubble burst and the financial system imploded?

    Did that many people decide to become voluntarily unemployed?

    1. I'd say they made the best time allocations they could, given their rapidly deteriorating circumstances. You have a problem with that?

    2. I'm with Peter here. And yes, I have a problem with that. The school leavers in 2009 are like puppies in a pet shop waiting for some adoring family to give them a good home. Six years later they are yesterday's goods.

    3. Alright, Roger, so you're claiming that the unemployed did not make the best time allocation decisions they could, given their rapidly deteriorating circumstances? Or you are claiming they had no choices (you still haven't answered the spike in exit rates at points where UI benefits expire, btw). Well, I guess we can agree to disagree here. (And before you agree with Peter too wholeheartedly, read his next comment.)

      One other thing to point out: a non-employed person "waiting" for a job is not counted as unemployed by the LFS. That measure of unemployment you have on your blog does not count people out of the labor force (i.e., discouraged workers).

    4. The spike you cite is entirely consistent with people voluntarily choosing to leave the labor force.

      And as for discouraged workers. Those adorable puppies are the unemployed. And they are searching! Those long languid looks they give when you peek into the cage require effort.

  4. It's amazing to me that economists can differ on this subject. Again I believe the culprit must be the corruption of money and ideological thinking. As Upton Sinclair wrote "“It is difficult to get a man to understand something, when his salary depends on his not understanding it."

    If unemployment is voluntary, why should the government try to get unemployment levels down via monetary or fiscal policy? Why have unemployment insurance? Why have a welfare state or safety net?

    I just saw the Ivan Reitman movie Dave the other day:

    1. "If unemployment is voluntary, why should the government try to get unemployment levels down via monetary or fiscal policy? Why have unemployment insurance? Why have a welfare state or safety net?"

      I addressed that question in my post. Perhaps you read it through too quickly. Roger has constructed models where all individual actions are privately optimal, voluntary, rational, whatever you want to call it. Nevertheless, it is possible that collectively these actions result in a suboptimal allocation calling for government intervention.

      Please read this, it might help:

    2. Sometimes language matters. I want to reintroduce 'involuntary' into the language because it reminds us of the point that David makes in his comment; equilibria are not always optimal. Personally, I would go much further. In the absence of government stabilization policies; equilibria are almost always very far from being optimal.

    3. I agree, Roger, sometimes language matters. I think it matters a lot in the way labor market policies are designed (this is where I'm coming from). You are coming at it from more of a macro perspective--you and others would like to remind us that recessions are not optimal. I don't think many people are going to disagree with you, even if you drop the "involuntary unemployment" label.

      But OK...most social games have multiple equilibria that are generically suboptimal. The world is full of involuntary consumption, saving, unemployment, leisure, etc. Again, I know what you're saying, I just don't find the language useful.

      I'm glad that we finally found something to disagree on! :)

  5. This comment has been removed by the author.

  6. i. go to bar - state we can describe as voluntary
    ii. get turned down at the bar - state we can describe as involuntary
    iii. turn someone down at the bar - state we describe as voluntary
    Don't really see the problem with this language ... It is essential to how we discuss the notion of free will

  7. I believe that Mr. Andolfatto needs to be fired. The experience will be enlightening for him.

    1. Dr. Morbius:

      As I mentioned in my blog post (which I suspect you have not read), I had the unhappy experience of being laid off from my construction job in December 1981. I was lucky at the time because I was young and had no family to support. I think it's fairly easy for me to imagine how much worse things could have been.

      Again, if you read my post carefully, you will see that I take pains that the labels economists attach to labor market choices should not have any bearing on how we (or policy) should intervene to help people who find themselves in difficult circumstances.

  8. The same can be said for many people "not in the labor force".

    Most People don’t Leave the Labor Force: Through no fault of their own, many are brutally shoved out.

  9. David may believe the terms "voluntary" and "involuntary" may be unhelpful in the meaning of those terms used in his DSGE models.

    Possibly - but so much the worse for his DSGE models. The distinction is all too useful when used in the everyday sense of the word, which is what those seeking to destroy the welfare state, further skew the distribution of wealth and income to the rentier class and then blame the victims of those policies for their fate.

    David, the welfare losses from a recession are disproportionately concentrated upon those least able to bear them - something a pathetically inadequate representative agent models cannot capture in its nature. Your devotion to a modelling approach that DELIBERATELY abstracts from that is precisely why so many people see that devotion as simply clever shilling for those rentiers.

    Oh, and on the transition for UI issue read Atkinson and Micklewright's 1991 JEL review article, which IMO comprehensively demolishes that approach as a way to measure the elasticity of unemployment WRT benefit replacement rates.

    1. Derridaderider:

      It is true that representative agent models cannot capture the nature of the phenomenon. But where in my argument have I employed a representative agent model? Again, you, like many others, like to make moralizing statements based without reading what I had to say. In my post I *explicitly* mention heterogeneity, as does the article I linked to.

      So, no, my approach does NOT deliberately abstract from the fact that hardship is concentrated on the unemployed and working poor. I would say, however, that you deliberately did not read my post, or my article, preferring instead to just presume what I said and pontificate from some imagined higher moral ground.

  10. Is it possible that what David Andolfatto is trying to say is that use of the phrase, "involuntary unemployment," is redundant? That unemployment is always involuntary? If that's what he's trying to say I would agree with him. If he's trying to say that unemployment, especially long-term unemployment, is something people choose because it maximizes their utility, I think that's idiotic.

    1. Roger,

      Yes, I think the term is redundant in a sense, but probably not for the reason you think.

      I think of individuals circumstances as having an involuntary component. People cannot control, nor do they wish for, terrible life-changing circumstances.

      BUT, given these circumstances, and given competing uses of time, people *do* choose to allocate their time across these competing uses. And sometimes, given circumstances (e.g., given a source of nonlabor income), it makes sense to spend time looking for a good job, rather than wasting time at a lousy job.

      Consider what might happen to such a person if they did not have UI benefits. They might instead choose to work at the first crappy job they find. Would you as a policymaker be pleased with such an outcome. After all, employment goes up and "involuntary" unemployment goes down.

      That is nonsense, in my opinion.

    2. He's saying people can always find a lousy job.

      Here's Piketty on Europe where a lot of people are voluntarily unemployed.

      "Piketty: It's time for us to think about the young generation of Europeans. For many of them, it is extremely difficult to find work at all. "

      It's economics by moralism.

    3. Goodness, Peter, I honestly do not think you read what I wrote. So what if people can always find a lousy job? That doesn't mean they should choose it. And, indeed, people are sometimes compelled to choose lousy jobs (over spells of unemployment). Morality has nothing to do with these observations.

  11. Two points:

    First, certainly sometimes a person cannot get a job where they are paid -- no matter how poorly compensated -- if they want one. They or members of their family may even starve to death as a consequence. If that situation doesn't count as "involuntary unemployment" then I really don't understand how we are using the term.

    If we do admit this occurs in some cases then we are just haggling about the boundary.

    Second, reading David's post closely, investment in search gets excluded early on because it doesn't fit into neoclassical models. This is clearly assuming the conclusion. Later on he mentions search but sets that aside. Then he claims that readers who disagree with him about involuntary unemployment don't think incentives matter, and also claims that "involuntary" is just used to emphasize that many unemployed "are not typically happy" (which, he says, is also true of many who are employed).

    Somehow the whole question of whether someone can get a job has gotten lost in the shuffle. However I don't see an argument here. Whether neoclassical models can model the status of involuntary unemployment is irrelevant to whether that category is a valid description of social reality. He makes no argument to support his claim that describing some conditions as "involuntary unemployment" implies one doesn't believe in incentives.

    All this is unfortunate because apparently Andolfatto does know about the current state of search research, could discuss unemployment with job search as investment and maybe could have sharpened up our discussion of people who want a paying job but would say that they can't find one. But this opportunity was completely bypassed in favor of muddled policing of the word "involuntary".

  12. Roger, you are correct that, in the General Theory, Keynes defines involuntary unemployment as a situation where the MRS between consumption and leisure is different from the wage. However, Keynes's starting point was a Walrasian market, so the implication of his definition is that there is a shortage of jobs. This is exactly what most people think today when they use the term "involuntary"; a wage stuck above a Walrasian equilibrium resulting in more people willing to work than jobs available. This is very different from what goes on in search models. Yes, the MRS is still different from the wage, but for reasons that Keynes ignored or in fact, if I remember correctly, he dismissed: Doesn't he say that his involuntary unemployment is in addition to the frictional unemployment, which is actually the type of unemployment generated by costly search?


Note: Only a member of this blog may post a comment.